Algorithmic Justice Or Algorithmic Bias? Examining The Constitutional Limits Of Artificial Intelligence In India’s Criminal Justice System
Aryan Verma, LL.B. (Hons.) University of Allahabad
ABSTRACT
Facial recognition cameras trained on protest sites, a predictive-policing dashboard colouring Delhi’s neighbourhoods by projected crime risk, and a High Court bail order that pauses, mid-reasoning, to consult a chatbot India’s criminal justice system has begun to acquire artificial intelligence one procurement contract at a time, without a single statute drafted with algorithms in mind. This paper asks whether that acquisition can survive Articles 14 and 21 of the Constitution and the common-law principles of natural justice that give those Articles their operative content. It argues that the doctrinal materials for an answer already exist the anti-arbitrariness principle of Royappa and Maneka Gandhi, the proportionality standard of Puttaswamy, the reviewability principle of Anuradha Bhasin, and the anti- vagueness doctrine of Shreya Singhal but that no case has yet welded them into a standard capable of being applied to a machine-generated output. Reading these strands together with the EU AI Act’s risk-based architecture and the cautionary lessons of State v. Loomis and Carpenter v. United States, the paper proposes a four-part constitutional floor transparency, reviewability, accountability and contestability, converging on an enforceable right to explanation that must be satisfied before any AI- generated output is permitted to influence a person’s liberty. Its central and, it hopes, its principal claim to originality is this: India does not yet need to resolve whether AI should be used in criminal justice; it needs, first and as a matter of constitutional obligation rather than policy preference, a framework for Explainable AI.
Keywords: Artificial intelligence, Article 14, Article 21, natural justice, facial recognition, predictive policing, algorithmic due process, right to explanation, Explainable AI, EU AI Act, criminal justice, privacy.
